Contribution of Trabecular Microarchitecture and its Heterogeneity
to Biomechanical Behavior of Human L3 Vertebrae
Wegrzyn, J; 1Roux, J P;
1Arlot, M E;
1 Vilayphiou, N;
1Delmas, P D;
3Bouxsein, M L
+1INSERM U 831, Universit de Lyon, Lyon, France,
2Department of Orthopedic Surgery, Pavillon T, Hpital E. Herriot, Lyon, France,
3Orthopedic Biomechanics Laboratory, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA
Senior author [email protected]
INTRODUCTION: Low bone mineral density (BMD) is a strong risk factor for
vertebral fracture in osteoporosis. However, BMD explains only
40 to 70% of the variation in trabecular bone strength . In
addition, many fractures occur in people with normal BMD .
Besides BMD, trabecular microarchitecture improves prediction
of bone mechanical behavior [3, 4]. Trabecular microarchitecture
heterogeneity has been previously described [5, 6], however
there is limited information about its contribution to vertebral
The aim of this study was to assess the contribution of
trabecular microarchitecture and its heterogeneity to mechanical
behavior of human lumbar vertebrae.
METHODS: L3 vertebrae taken from 21 fresh donors (11 males and 10
females, respectively aged 7510 years and 7610 years) were
analyzed and destructively tested in uniaxial compression.
Lateral-BMD (g/cm) was measured using dual energy X-ray
absorptiometry (DXA; Delphi W, Hologic, MA, USA). 3D
trabecular microachitecture (bone volume per tissue volume
(BV/TV), structural model index (SMI), trabecular separation
(Tb.Sp*), trabecular thickness (Tb.Th*) and trabecular number
(Tb.N*)) was assessed without model assumption using high-
resolution peripheral quantitative computed tomography (HR-
pQCT; X-Trem CT, Scanco Medical, Switzerland) with a
nominal isotropic voxel size of 82 m.
Trabecular microarchitecture heterogeneity was assessed
using two 8.2 mm diameter virtual biopsies (one anterior and one
posterior vertically cored in 3 zones: superior, middle and
inferior) and on the whole vertebral trabecular area (Figure 1.).
Heterogeneity of trabecular microarchitecture was evaluated by:
1) the ratio of antero-posterior BV/TV (BV/TVratio); 2) the
coefficient of variation of the vertical 3 zones parameters
(BV/TVcv); and 3) the distribution expressed by the standard
deviation of Tb.Sp* on virtual biopsies (Tb.Sp*SDant) and on the
whole trabecular area (Tb.Sp*SD).
Vertebral stiffness (N/mm), failure load (N) and work to
failure (N.mm) were measured on the whole vertebral body
using a servohydraulic testing machine (Schenck RSA-250,
Germany) under displacement control at 0.5 mm/s until failure.
All parameters had normal distribution, after log
transformation for work to failure and BV/TVratio. Parametric
tests (Student t-test, ANOVA, Pearson's correlation coefficient,
stepwise and multiple regression analyses) were performed using
SPSS 12.0 software.
RESULTS: Mean values for L3 vertebral biomechanics, as well as
trabecular architectural parameters are given in Table 1.
Microarchitectureal features differed significantly between
the anterior and posterior virtual biopsies for all parameters (p=
0.007 to 0.04), except for SMI and Tb.Th*. Vertebral failure
load was mainly explained by microarchitectural parameters of
the anterior region as assessed by stepwise regression analysis
(ie.; in the equations, failure load = anterior and posterior
microarchitectural parameter, the second one was always out of
model). As a result, we studied vertical heterogeneity on the
anterior biopsy. Significant vertical heterogeneity was found for
BV/TV and SMI (respectively, p-value = 0.0001 and 0.021,
Although all parameters reflecting trabecular
microarchitecture of the whole vertebral body (i.e.; BV/TV,
SMI, Tb.Sp*, Tb.Th* and Tb.N*) were correlated with
mechanical indices (r = -0.81 to 0.44, p-values = 0.0001 to
0.047, Pearson test), SMI seemed to be the most pertinent one,
with r = -0.81 (stepwise regression analysis).
Trabecular bone heterogeneity (BV/TVratio and Tb.Sp*SDant)
was significantly correlated with mechanical behavior (r = -0.53
to -0.57, p = 0.01 to 0.007). No correlation was found with
The combination of one parameter of bone mass (BMD) and
one parameter of heterogeneity (BV/TVcv) significantly
improved the prediction of both failure load and stiffness as
indicated by stepwise multiple regression analyses (respectively,
R = 0.83 and 0.74; p